School Performance and Network Effects among Classmates

Emma Zavarrone, Università IULM
Agnese Vitali, Università Bocconi

This paper contributes to the literature on the influence of network structure on performance of university students. The paper compares the marginal effect on performance of five different aspects of communication among classmates, represented by five network structures: study, friendship, exchange of information, exchange of lecture notes, and trust networks. Using a set of different weight matrix specifications based on adjacency, outdegree, cohesion, reciprocity and homophily criteria, we show that members of a communication network tend to have similar performances. Results, however, are not always robust to different weight matrix specifications. We conclude that in the study of social interaction processes the network structure need to be translated into a meaningful and theory-guided weight matrix specification.